"Turbo" codes offer to cut radio net power use

RESEARCH TRIANGLE PARK, N.C. - Battlefield radio system engineers are pushing advanced software design to tackle one of their most vexing problems: the high power consumption of mobile radio networks.

By Wilson Dizard III

RESEARCH TRIANGLE PARK, N.C. - Battlefield radio system engineers are pushing advanced software design to tackle one of their most vexing problems: the high power consumption of mobile radio networks.

Working with funding from the Army Research Office in Research Triangle Park, N.C., university researchers are determining how to cut power consumption in mobile radio networks that must provide reliable, anti-jam communications for as long as three full days.

Battlefield sensor platforms such as unmanned aerial vehicles, as well as individual soldiers, rely on batteries on increasingly long missions, explains James Freebersyser, program manager for the Army Research Office in Research Triangle Park, N.C.

"The problem is, batteries aren`t getting better," Freebersyser admits. The answer, he says, is not merely effecting a quantum leap in the quality and longevity of batteries. "We think we can get orders of magnitude improvements, however, by taking a systems approach."

Electronics that run on batteries are fundamentally different in commercial and military applications, Freebersyser points out. Commercial electronics normally run on batteries for 18 hours before recharging. Military electronics, however, must run for 72 hours between battery recharges - four times as long as commercial systems.

Freebersyser explains that diverse teams of engineers, each working on one segment of a network, carried out previous efforts to reduce radio network power. "That usually meant, making one subsystem at a time for low-power," he says. "So everybody makes their little world low power, but shifts the power demand to another little world. For example, you could develop exotic coding techniques at the transmitter. But that might increase the power needed for decoding at the receiver end."

Army Research Office officials sought to lick the problem by commissioning basic research under a DOD Multidisciplinary University Research Initiative (MURI) on Low Energy Electronics Design for Mobile Platforms. Specialists at the Electrical Engineering and Computer Science Department at the University of Michigan at Ann Arbor are carrying out the work under supervision of Principal Investigator Wayne Stark.

Some aspects of the project, especially the turbo coding research, already have attracted attention from officials ITT Corp. in Fort Wayne, Ind., who are looking at ways to integrate the technology into the next upgrade of the Single Channel Ground-Air Radio System, better known as SINCGARS, he says.

The MURI project has broken down the low-energy electronics design problem into four layers.

First, at the distributed system layer, researchers are asking, who should communicate with whom and when? What information should be exchanged in real time among mobile platforms and their commanders? How can operators maintain the connectivity of the network with minimal energy consumption? The researchers seek to achieve a reliable, adaptable, deadlock-free reconfigurable communications net that uses less power and energy.

Second, at the local integration layer, questions focus on the design of radio systems for individual platforms, integration of subsystems, and compliance with size, weight, and reliability constraints while slashing power and energy consumption.

Third, at the processing layer, the researchers are probing signal and information processing algorithms including adaptive coding and modulation for wireless transmission; data compression for voice, image, and video; and design of efficient decoders.

Finally, at the device layer, the MURI researchers are developing models of chips, circuits, and antennas to implement the algorithms developed at the processing layer.

At the end of the 5-year project, Stark and his team members plan to pull together a simulation of a mobile platform that will optimize the design goals. The MURI project receives $1 million annually for each of its five years, and has been in progress for about 18 months.

"A communications net is a very complicated thing," Stark says. "You have modulation, coding and network protocols. It takes quite a wide variety of people to design the whole thing. In the past, they have operated independently. In our multidisciplinary effort, we`re looking at things like micro-electro-mechanical switches. The overall question is, how do you adapt modulation, coding, and multiple access technologies to have extremely low power consumption systems? What we have to do is rely on joint simulation and optimization. We have to build up modules, and subsystems, and analyze [power consumption and other] tradeoffs."

Stark says he recently had visited the Army`s Communications-Electronics Command at Fort Monmouth, N.J., to describe the work of the MURI team and keep in touch with the Army`s needs. A MURI researcher will be assigned for long-term liaison with CECOM radio design specialists, he added.

One of the most promising techniques developed under the MURI program is the turbo coding technique for spread-spectrum radios. Turbo codes have been known for about five years, and are based on a mathematical technique developed in France, Stark says.

"They were shown to have excellent performance in benign environments," Stark says, referring to the low-interference, no-jamming world of commercial radio. "We`re looking at this for frequency-hopping radios. We modified the algorithms from the original design so they would work with frequency-hopping radios."

The key advantage of using the turbo coding technique is a transmitted power reduction of 6 decibels (dB), Stark says. "It can handle interference in a better way than other coding techniques so you don`t need as much power."

Turbo coding introduces redundancy into the data stream of a radio message, using interleaved signals. "The decoding algorithm is where the turbo code comes in," Stark explains. "You have two different decoders operating. One estimates the data first and then passes it on to the second. Turbo coding is for error control coding to achieve more reliable communication."

The central difference between turbo codes and conventional codes is a mathematical feature of the algorithm. "The traditional goal used in code design is to maximize the minimum distance between codewords," Stark says. "This gives good performance at high signal-to-noise ratios."

In turbo coding, by contrast, the goal is to minimize the average number of codewords at each distance. This involves a disadvantageous sacrifice in the minimum distance between codewords, leading to a smaller minimum distance between codewords. But the greatly offsetting advantage is the very few codewords at the minimum distance.

Stark and his fellow investigators are passing on their results to the ITT SINCGARS team, and working to refine the turbo coding algorithms. An immediate goal is to reduce the complexity of the decoding function. They also are investigating the performance of the system using different packet lengths.

Stark and others from the MURI project will describe some of their results at the upcoming MILCOM `98 conference in Boston in October. Look for further information about the Army Research Office MURI project results on the Internet at http://www.eecs.umich.edu/aromuri96/ index.html.

More in Power